  Rank: Advanced Member Groups: Registered
Joined: 28/08/2014(UTC) Posts: 1,478  Was thanked: 930 time(s) in 590 post(s)
|
Originally Posted by: grelkin2  ... drawbacks. In linux smath does not work very well. I have all calculations going on the right side of de system. I calculate the coordinates of the tangent point and respectively the rotation speed of the oval. At the tangent point the velocity is zero and a tick icon is observed. ... Hi grelkin2. I don't have a linux rigth now, and can tell which issue could be that. About the numeric precission, there are two big sorces of numerical errors. First, the arc length, which was "calculated" by eye looking at the graph when the curve closes, given the value of 7.3. This value can be improved numerically by solving an equation that sets the square of the norm between the final and initial points equal to zero. This problem is not difficult to solve. As comment, this is why I notate in other posts under this topic that one need to divide with the norme of the Dragile system to parametrize about the arc lenght. The second major source of error is considering the minimum of the entire curve only by taking into account a few points: only 31 for each rotated figure, from which the minimum is selected at each iteration. It could be improved by considering many more points, but it would be very slow and impractical, so it would be necessary to interpolate assuming a smooth curve, but it would require many steps to achieve this. Finally, rkfixed is used, which is the least accurate of the wide variety of ode solvers available. Best regards. Alvaro.
|
|
|
|
|
|
 Rank: Advanced Member Groups: Registered
Joined: 12/12/2020(UTC)
Posts: 43
Was thanked: 5 time(s) in 5 post(s)
|
Originally Posted by: Razonar Hi grelkin2. I don't have a linux rigth now, and can tell which issue could be that. About the numeric precission, there are two big sorces of numerical errors. First, the arc length, which was "calculated" by eye looking at the graph when the curve closes, given the value of 7.3. This value can be improved numerically by solving an equation that sets the square of the norm between the final and initial points equal to zero. This problem is not difficult to solve.
As comment, this is why I notate in other posts under this topic that one need to divide with the norme of the Dragile system to parametrize about the arc lenght.
The second major source of error is considering the minimum of the entire curve only by taking into account a few points: only 31 for each rotated figure, from which the minimum is selected at each iteration. It could be improved by considering many more points, but it would be very slow and impractical, so it would be necessary to interpolate assuming a smooth curve, but it would require many steps to achieve this.
Finally, rkfixed is used, which is the least accurate of the wide variety of ode solvers available.
Best regards.
Alvaro. Arc length is about 7.335. I changed N is your sm file to 121 points for smooth curves. rkfixed is a reliable basic method, the other methods are needed for optimization or experimentation or have specific uses. Below is a plot of the translational velocity of the oval at which the velocity at the tangent point is zero. 
|
|
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 28/08/2014(UTC) Posts: 1,478 Was thanked: 930 time(s) in 590 post(s)
|
Hi grelkin2 Originally Posted by: grelkin2 Arc length is about 7.335. ... This is how Dragilev's method can be used to calculate the arc length of a closed curve numerically. My improvement on Dragilev's method consists in dividing the system of differential equations by its norm to obtain the parameterization of the curves along their arc length. I have not seen this observation anywhere else.   Arc length of a planar closed curve - Dragilev Method.sm (30kb) downloaded 2 time(s).Originally Posted by: grelkin2 ... I changed N is your sm file to 121 points for smooth curves. rkfixed is a reliable basic method, the other methods are needed for optimization or experimentation or have specific uses. ... If you say ... Originally Posted by: grelkin2 Below is a plot of the translational velocity of the oval at which the velocity at the tangent point is zero. Yes, like my approach with 31 points, except that your zero is more zero than mine, which is only an approximation, but a pretty good one, because to detect it you have to zoom in on the graph. There is no numerical method that can withstand a sufficiently large zoom on its graph. It looks like you use more than 3000 points, something SMath couldn't handle in this case. One question: what's the interpration of the values of the x-axis? Best regards. Alvaro.
|
|
|
|
|
|
 Rank: Advanced Member Groups: Registered, Advanced Member Joined: 10/11/2010(UTC) Posts: 1,620   Was thanked: 1382 time(s) in 806 post(s)
|
You can configure the method using a rkfixed variable. rkfixed := "RK1", …, rkfixed := “RK6”. In Mathcad, it seems to me this method with a variable step. When I figure it out as this is done, I will make an update. Edited by user 31 October 2024 18:28:53(UTC)
| Reason: Not specified |
Russia ☭ forever (git.io/vOZo0)
Viacheslav N. Mezentsev |
|
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 12/12/2020(UTC)
Posts: 43
Was thanked: 5 time(s) in 5 post(s)
|
Originally Posted by: uni You can configure the method using a rkfixed variable. rkfixed := "RK1", …, rkfixed := “RK6”.
In Mathcad, it seems to me this method with a variable step. When I figure it out as this is done, I will make an update. Будет замечательно если ты подготовишь модель и рассмотришь зависимость времени вычисления от порядка при заданной точности. It will be great if you prepare a model and consider the dependence of the calculation time on the order at a given accuracy.
|
|
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 12/12/2020(UTC)
Posts: 43
Was thanked: 5 time(s) in 5 post(s)
|
Originally Posted by: Razonar
Yes, like my approach with 31 points, except that your zero is more zero than mine, which is only an approximation, but a pretty good one, because to detect it you have to zoom in on the graph. There is no numerical method that can withstand a sufficiently large zoom on its graph. It looks like you use more than 3000 points, something SMath couldn't handle in this case. One question: what's the interpration of the values of the x-axis?
Best regards.
Alvaro.
You could have sufficiently large zoom in octave. olv2.mov (162kb) downloaded 2 time(s).Value of the x-axis is the number of points at a given accuracy. I changed the accuracy and, accordingly, the number and combined everything in one animation. olv3.mov (366kb) downloaded 2 time(s).
|
|
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 28/08/2014(UTC) Posts: 1,478 Was thanked: 930 time(s) in 590 post(s)
|
|
|
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 12/12/2020(UTC)
Posts: 43
Was thanked: 5 time(s) in 5 post(s)
|
Originally Posted by: Razonar   Best regards. Alvaro. [RUS]В этих замечательных анимациях можно наблюдать движение замкнутых кривых трех видов в зависимости от соотношения скорости движения центра и скорости вращения в точке касания. Движение с пробуксовкой, движение без проскальзывания и движение с проскальзыванием получаются при скорости движения центра меньше скорости вращения в точке касания, равно и больше соотвественно. Владея методом Драгилева можно задавать любой закон движения. Ниже анимация с ускорением скорости движения, при которой можно наблюдать плавный переход между тремя видами движения, описываемых раннее. Также в эту анимацию включил движение без проскальзывания с циклоидоподобными кривыми.[/RUS] [ENG]In these wonderful animations, you can observe the movement of closed curves of three types, depending on the ratio of the speed of movement of the center and the speed of rotation at the point of contact. Movement with stalling, movement without slippage and movement with slippage are obtained when the speed of movement of the center is less than the speed of rotation at the point of contact, equally and more respectively. Using the Draghilev method, you can set any law of motion. Below is an animation with an acceleration of the speed of movement, at which you can observe a smooth transition between the three types of movement described earlier. Also included in this animation is a non-slip motion with cycloid-like curves.[/ENG]  olvm.mov (1,246kb) downloaded 0 time(s).
|
|
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 28/08/2014(UTC) Posts: 1,478 Was thanked: 930 time(s) in 590 post(s)
|
|
|
|
|
|
|
 Rank: Advanced Member Groups: Registered
Joined: 30/09/2012(UTC) Posts: 64 Was thanked: 13 time(s) in 8 post(s)
|
Yesterday I asked the author of one publication, which mentioned the use of the Method when constructing graphs in Mathcad, to write and publish a scientific paper in a peer-reviewed journal. 11`22`Edited by user 09 November 2024 08:59:47(UTC)
| Reason: Not specified
|
|
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 12/12/2020(UTC)
Posts: 43
Was thanked: 5 time(s) in 5 post(s)
|
  [RUS]Движение эллипса без проскальзывания с поворотом относительно эллипса. Анимация грубая, погрешности большие. Решение получил не точное из системы ду, а комбинированное(сумма нескольких движений)[/RUS] [ENG]IMovement of the ellipse without slippage with rotation relative to the ellipse. The animation is rough, the errors are large. I got an inaccurate solution from the system of de's, but a combined one (the sum of several movements)[/ENG] elnel.mov (1,862kb) downloaded 2 time(s).
|
|
|
|
|
|
Forum Jump
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.
